The research will apply information processing methods to processes that underlie the initial stages of reading. It will focus on early stages, i.e., visual input of letters, words, and nonsense strings of written Serbo-Croatian. Emphasis will be on units of recoding from visual to phonetic form and linguistic level(s) where visual and speech processes merge. There are special advantages for comparisons with English, since Serbo-Coatian offers exceptional opportunities in visual information processing: (1) there is a single spoken laguage, but two alphabets (Cyrillic and Roman) which are used interchangeably by substantial parts of the population, (2) the script systems are regularly phonemic, with one-to-one correspondence of letter to sound, (3) some letter shapes (about a third) are common to the two alphabets, and of these, some have the same sounds, but some do not, and (4) schooling introduces both alphabets at an early age, with Cyrillic taught first in some areas, and Roman first in other areas. Initial work with adult readers will study their visual encoding processes; extensions are planned to developmental studies of reading acquisition by children. (The research is to be done in close cooperation with the Haskins Laboratories.)